OpenAI and Anthropic are projected to buy or lease 26 million Google TPUs in 2028-29, while Alibaba unveiled Qwen4-Max with a million-token context window. Naver and Kakao joined the personal AI agent race, and Samsung, SK and LG are tightening hiring supervision against AI-powered cheating.
🔮 OpenAI, Anthropic to Buy 26M Google TPUs in 2028-29, Kuo Projects
TF International analyst Ming-Chi Kuo said on November 16 that his industry survey points to OpenAI taking about 12 million Google eighth-generation TPUs and Anthropic about 14 million across 2028 and 2029. The two labs' demand would amount to 79-87% of his total TPU shipment estimate for those years, extending Google's TPU shipment visibility to 2029, ahead of Nvidia GPUs and other AI accelerators.
🧠 Alibaba Unveils Qwen4-Max… Million-Token Context and Native Vision
Alibaba released its flagship model Qwen4-Max, featuring native vision, a million-token context window, professional tool use and sustained execution of complex tasks. Tencent also shipped Hunyuan 4 Flash the same day, a 224-billion-parameter model with a million-token context, intensifying competition among Chinese open-model makers.
💳 OpenAI Launches ChatGPT Wallet… Fingerprint Approval for Every Agent Purchase
OpenAI launched ChatGPT Wallet, letting agents complete purchases directly. Every purchase requires Face ID or fingerprint approval, so agents never receive users' real card details. Meanwhile, United Airlines is blocking outside AI agents from its booking systems, opening a fight over who controls agent-driven payments and reservations.
🤖 Anthropic Launches Claude Meetings… Fable 5.2 Builds an OS with 1,000 Agents
Anthropic launched Claude Meetings, which generates editable meeting notes, action items, whiteboards and spreadsheets in real time and archives them automatically. Whiteboard, spreadsheet and data analysis tools arrived alongside it, with more than 9,000 connectors in its marketplace. In a test by Wharton professor Ethan Mollick, Fable 5.2 led over 1,000 agents to build a bootable operating system in eight days and formalize Fermat's Last Theorem in 30 hours.
⚖️ AI Safety Lawsuit Seeks Development Freeze at Five Companies… Court Ruling Key
On November 12, four nonprofit groups sued OpenAI, Google, Anthropic, Meta and SpaceX's AI division in a San Francisco court, seeking limits on the development and deployment of frontier AI systems over public-safety and loss-of-control risks. No court order halting development has been issued, and whether the court accepts the injunction request will be decisive.
🇰🇷 Naver, Kakao Join the Personal AI Agent Race… Naver Plans Launch This Year
Following global big tech, Naver and Kakao are stepping into the personal AI agent race. Naver plans to launch a personal AI agent within the year that handles shopping, reservations and payments on behalf of users, while Kakao is expanding its KakaoTalk-centered AI agent business, linking it with advertising and commerce and targeting break-even in its AI business in 2028. Korean cloud companies including Naver, KT and NHN are also accelerating data center expansion on AI demand.
🎓 Samsung, SK, LG Scramble Over AI Cheating… Smart Glasses Banned, Wearable Checks Tightened
After the top scorer of a tax accountant exam turned himself in for cheating with AI, major Korean conglomerates including Samsung, SK and LG are scrambling to counter AI-assisted cheating in hiring. Samsung has added a smart-glasses ban to its applicant manual, and SK is strengthening wearable-device checks and online exam supervision. Smart glasses and tiny cameras combined with generative AI are making cheating hard to detect with traditional supervision alone.